CN112761628B - Shale gas yield determination method and device based on long-term and short-term memory neural network - Google Patents
Shale gas yield determination method and device based on long-term and short-term memory neural network Download PDFInfo
- Publication number
- CN112761628B CN112761628B CN202110096222.8A CN202110096222A CN112761628B CN 112761628 B CN112761628 B CN 112761628B CN 202110096222 A CN202110096222 A CN 202110096222A CN 112761628 B CN112761628 B CN 112761628B
- Authority
- CN
- China
- Prior art keywords
- target
- yield
- shale gas
- data set
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH DRILLING; MINING
- E21B—EARTH DRILLING, e.g. DEEP DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B49/00—Testing the nature of borehole walls; Formation testing; Methods or apparatus for obtaining samples of soil or well fluids, specially adapted to earth drilling or wells
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The embodiment of the specification provides a shale gas yield determination method and device based on a long-term and short-term memory neural network, wherein the method comprises the following steps: obtaining the yield data of the target shale gas horizontal well at each moment in the period before the target moment; preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time; according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment. In the embodiment of the specification, the influence of the yield of a plurality of moments before the target moment on the yield of the target moment is considered, and the accuracy of yield prediction is effectively improved.
Description
Technical Field
The embodiment of the specification relates to the technical field of shale gas exploration and development, in particular to a shale gas yield determination method and device based on a long-term and short-term memory neural network.
Background
In recent years, shale gas development is gradually becoming a new hot spot of world energy development, and shale gas mainly exists in shale rich in organic matters and interlayers and exists in adsorbed gas and free gas. The shale gas exploration and development has important social and economic significance, the shale gas productivity prediction can be accurately carried out, and the shale gas exploration and development guiding method has important guiding significance for guiding shale gas exploration and development.
In the prior art, the shale gas productivity is generally predicted by adopting an analytical formula method, wherein the analytical formula method is to deduce an analytical solution of a shale gas productivity formula by establishing a mathematical model. Before the establishment of the mathematical model, however, the ideal physical model assumption must be relied on, so that the limitation to the shale gas well is more, and the established mathematical model is only a theoretical model and has larger deviation from the actual situation. Therefore, the analytic formula method in the prior art is low in applicability, and the shale gas productivity cannot be accurately predicted.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the specification provides a shale gas yield determination method and device based on a long-term and short-term memory neural network, and aims to solve the problem that the yield of shale gas cannot be accurately predicted in the prior art.
The embodiment of the specification provides a shale gas yield determination method based on a long-term and short-term memory neural network, which comprises the following steps: obtaining the yield data of the target shale gas horizontal well at each moment in a period before the target moment; preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time; according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-short term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment.
The embodiment of the present specification further provides a shale gas production determining apparatus based on the long-term and short-term memory neural network, including: the acquisition module is used for acquiring the yield data of the target shale gas horizontal well at each moment in the period before the target moment; the preprocessing module is used for preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set includes yield data at a plurality of time instants arranged in time; the determining module is used for determining the yield of the target shale gas horizontal well at the target moment by utilizing a yield prediction model obtained by pre-training according to the target yield data set; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment.
The embodiment of the specification further provides shale gas production determination equipment based on the long-short term memory neural network, which comprises a processor and a memory, wherein the memory is used for storing processor executable instructions, and the processor executes the instructions to realize the steps of the shale gas production determination equipment based on the long-short term memory neural network.
The present specification also provides a computer readable storage medium having stored thereon computer instructions which, when executed, implement the steps of the long-short term memory neural network-based shale gas production determination method.
The embodiment of the specification provides a shale gas yield determination method based on a long-term and short-term memory neural network, which can obtain a target yield data set by acquiring yield data of a target shale gas horizontal well at each moment in a period before a target moment and preprocessing the yield data at each moment in the period before the target moment. Wherein the production data for a plurality of time instants in the target production data set may be arranged in time. Because the production is time-varying, the production at the target time is not independent, and therefore, the production of the target shale gas horizontal well at the target time can be determined according to the target production data set by using a production prediction model obtained by pre-training. The yield prediction model can be obtained based on long-term and short-term memory neural network training, and can be used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment. Therefore, the influence of the yield of a plurality of moments before the target moment on the yield of the target moment is considered, and the accuracy of yield prediction is effectively improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the disclosure, are incorporated in and constitute a part of this specification, and are not intended to limit the embodiments of the disclosure. In the drawings:
fig. 1 is a schematic step diagram of a shale gas production determination method based on a long-short term memory neural network according to an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of a shale gas reservoir numerical simulation model provided in accordance with an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of a parameter tuning learning curve for parameters in a long-short term memory neural network provided in accordance with an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of a yield decrement curve plotted using a training data set, a test data set, and a validation data set in accordance with an embodiment of the present description;
fig. 5 is a schematic structural diagram of a shale gas production determination apparatus based on a long-short term memory neural network according to an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of a shale gas production determination apparatus based on a long-short term memory neural network according to an embodiment of the present disclosure.
Detailed Description
The principles and spirit of the embodiments of the present specification will be described with reference to a number of exemplary embodiments. It should be understood that these embodiments are presented merely to enable those skilled in the art to better understand and to implement the embodiments of the present description, and are not intended to limit the scope of the embodiments of the present description in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As will be appreciated by one skilled in the art, implementations of the embodiments of the present description may be embodied as a system, an apparatus, a method, or a computer program product. Therefore, the disclosure of the embodiments of the present specification can be embodied in the following forms: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.
Although the flow described below includes operations that occur in a particular order, it should be appreciated that the processes may include more or less operations that are performed sequentially or in parallel (e.g., using parallel processors or a multi-threaded environment).
Referring to fig. 1, the present embodiment may provide a shale gas yield determination method based on a long-term and short-term memory neural network. The shale gas production determination method based on the long-term and short-term memory neural network can be used for accurately predicting the production of shale gas horizontal wells. The shale gas production determination method based on the long-short term memory neural network can comprise the following steps.
S101: and obtaining the production data of the target shale gas horizontal well at each moment in the period before the target moment.
In this embodiment, production data of the target shale gas horizontal well at each time in a period before the target time may be obtained. The target shale gas horizontal well may be a shale gas horizontal well whose yield is to be predicted, and the previous period of the target time may be a plurality of times of a preset time step before the target, for example: target time t 11 The preset time step is 10, then each time in the cycle before the target time may include: t is t 1 、t 2 、t 3 、t 4 、t 5 、t 6 、t 7 、t 8 、t 9 、t 10 。
In this embodiment, the numerical simulation yield data of the target shale gas horizontal well at the last moment of the previous period of the target time may also be obtained, and the numerical simulation yield data of the target shale gas horizontal well at the last moment of the previous period of the target time and the yield data of each moment of the previous period of the target time are used as the input data of the yield prediction model.
In this embodiment, since data that may affect the yield at the target time is limited, or data that may significantly affect the yield at the target time is limited, the preset time step may be an optimal time step obtained by training the long-short term memory neural network, and the preset time step may be a positive integer greater than 0, such as 5 or 10, and may be determined according to actual conditions, which is not limited in the embodiments of the present specification.
In the present embodiment, the time may be any specific date, for example: in 2018, day 10 and 10, of course, the time may be a time in months or an hour, and may be determined specifically according to actual needs, which is not limited in the embodiments of the present specification.
In this embodiment, the manner of obtaining the production data of the target shale gas horizontal well at each time in the period before the target time may include: and pulling the shale gas horizontal well from a preset database, or receiving the output data of the target shale gas horizontal well at each moment in the period before the target moment, which are input by a user. It is understood that the sample data set may also be obtained in other possible manners, for example, the yield data of the target shale gas horizontal well at each time in the period before the target time is searched in the web page according to a certain search condition, which may be determined according to actual conditions, and this is not limited in this specification.
S102: preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set includes yield data for a plurality of time instants arranged in time.
In this embodiment, since the production data at each time in the cycle before the target time may not meet the input requirements of the production prediction model in format, the production data at each time in the cycle before the target time may be preprocessed to obtain the target production data set.
In this embodiment, the numerical simulation yield data of the target shale gas horizontal well at the last moment of the previous cycle of the target moment may be preprocessed. The target production data set may include: the method comprises the steps of obtaining yield data at a plurality of moments arranged according to time and numerical simulation yield data at the last moment in a period of each time step, wherein the numerical simulation yield data can be obtained by utilizing a shale gas reservoir numerical simulation model to carry out history fitting.
In this embodiment, the preprocessing of the production data at each time in the cycle prior to the target time may include: filling missing values, rearranging the output data of all moments according to a time sequence, and the like. Of course, the preprocessing method is not limited to the above examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the embodiments of the present disclosure, and all the functions and effects that are achieved by the preprocessing method are also within the scope of the embodiments of the present disclosure.
In one embodiment, preprocessing the production data at each time in the cycle prior to the target time to obtain a target production data set may include: converting the yield data of each moment in a period before the target moment into data conforming to the input format of the long-short term memory neural network according to a preset time step, and obtaining a target yield data set; and the time data contained in the previous period of the target time is equal to the preset time step.
In the present embodiment, since the long-short term memory neural network is a recurrent neural network, which requires that the input data be time-ordered, the yield data in the target yield data obtained after the preprocessing may be arranged in time series.
S103: according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-short term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment.
In the embodiment, since the yield varies with time and the yield at the target time is not independent, the yield of the target shale gas horizontal well at the target time can be determined by using a yield prediction model obtained by pre-training according to the target yield data set, wherein the yield prediction model can be obtained by training based on a long-short term memory neural network and is used for predicting the yield at the target time according to the yield data at a plurality of times before the target time, so that the influence of the yields at the plurality of times before the target time on the yield at the target time is effectively considered, and the prediction accuracy is improved.
In this embodiment, the Long Short-Term Memory neural network (LSTM) is a Recurrent Neural Network (RNN), which can solve the problems of gradient extinction and gradient explosion during the Long sequence training process, and the LSTM can have better performance in a longer sequence than a common RNN. The LSTM is a special cycle body structure, and a forgetting gate, an input gate and an output gate in the LSTM can select information.
From the above description, it can be seen that the embodiments of the present specification achieve the following technical effects: the target yield data set can be obtained by obtaining the yield data of the target shale gas horizontal well at each moment in the period before the target moment and preprocessing the yield data of the target shale gas horizontal well at each moment in the period before the target moment. Wherein the production data for a plurality of time instants in the target production data set may be arranged in time. Because the production is time-varying, the production at the target time is not independent, and therefore, the production of the target shale gas horizontal well at the target time can be determined according to the target production data set by using a production prediction model obtained by pre-training. The yield prediction model can be obtained based on long-short term memory neural network training and can be used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment. Therefore, the influence of the yield of a plurality of moments before the target moment on the yield of the target moment is considered, and the accuracy of yield prediction is effectively improved.
In one embodiment, before determining the production of the target shale gas horizontal well at the target time using the pre-trained production prediction model according to the target production data set, the method may further include: setting attribute parameters of the model, and establishing a shale gas reservoir numerical simulation model according to the attribute parameters; wherein, a shale gas horizontal well is arranged in the shale gas reservoir numerical simulation model. The historical yield data set of the target shale gas horizontal well can be obtained, and the historical yield data set of the target shale gas horizontal well is utilized to perform historical fitting on the shale gas reservoir numerical simulation model to obtain numerical simulation yield data of the target shale gas horizontal well in a preset time period. The training data set can be generated according to the historical yield data set of the target shale gas horizontal well and the numerical simulation yield data of the target shale gas horizontal well in a preset time period. Furthermore, the long-term and short-term memory neural network can be trained according to the training data set to obtain a target yield prediction model. The training of the LSTM neural network is further constrained by using the numerical simulation result of history fitting, and the accuracy of prediction is improved in the aspect of long-term prediction.
In the embodiment, the long-term and short-term memory neural network can be trained together according to the actual historical yield data of the target shale gas horizontal well and the numerical simulation data subjected to historical fitting, so that a target yield prediction model is obtained. When the yield prediction is performed at a future moment, the historical yield data and the numerical simulation result of the historical fitting of the yield prediction model can be combined to be used as the input of the target yield prediction model for training and prediction. A method driven by a physical mechanism and data in a combined mode is formed, and the accuracy of prediction can be improved for the shale gas which is a relatively complex unconventional seepage problem.
In the embodiment, since a large amount of sample data is required for model training, the long-term and short-term memory neural network can be trained by directly using actual yield history data when a large amount of actual yield history data exists. Under the condition that actual perennial data are insufficient, the training data can be acquired in a numerical simulation mode to ensure the accuracy of the training data. Of course, the training data may also be obtained by combining the two manners, which may be determined according to actual situations, and this is not limited in the embodiments of this specification.
In this embodiment, the attribute parameters of the model may be basic parameters of a shale gas reservoir numerical simulation model, and may include parameters representing the size of the model, configuration parameters, and the like. The characteristic parameters can be geological parameters and/or engineering parameters influencing the productivity of the shale gas horizontal well, and the yield of the shale gas horizontal well can be accurately simulated by using the characteristic parameters. By using shale gas reservoir numerical simulation software of the embedded discrete fracture model, the shale gas reservoir numerical simulation model can directly calculate yield data of a preset age limit. For example, the preset age limit is within 15 years from the time point of the shale gas horizontal well production, and the shale gas reservoir numerical simulation software based on the embedded discrete fracture model can be used for simulating and obtaining the daily yield data of the shale gas horizontal well within 15 years from the time point of the shale gas horizontal well production.
In this embodiment, attribute parameters of the model may be preset, and a shale gas reservoir numerical simulation model may be established according to the set attribute parameters, as shown in fig. 2, where the shale gas reservoir numerical simulation model includes a shale gas horizontal well, and PERMX (mD) is a grid permeability in the x direction, and mD is millidarcy, a unit of permeability. Furthermore, historical fitting can be performed on the shale gas reservoir numerical simulation model according to actual yield data in the historical yield data set of the target shale gas horizontal well, so that numerical simulation yield data of the target shale gas horizontal well in a preset time period can be obtained. The shale gas reservoir numerical simulation model can be used for simulating a real shale gas reservoir, attribute parameters can be set according to a typical shale gas reservoir, and also can be set according to a target shale gas horizontal well, and the attribute parameters can be determined according to actual conditions, and the shale gas reservoir numerical simulation model is not limited in the embodiment of the specification.
In the embodiment, when the shale gas reservoir numerical simulation model is established, special seepage mechanisms such as Langmuir (Langmuir) adsorption and desorption, knudsen diffusion, crack stress sensitivity, gas-water two-phase seepage and the like of shale can be considered, so that the established shale gas reservoir numerical simulation model better conforms to the characteristics of an actual shale gas reservoir. The attribute parameters of the model may include at least one of: grid size, number of grids, length of shale gas horizontal well in the horizontal direction, langmuir pressure, limiting adsorption concentration, critical desorption pressure, and pressure for constant pressure production. For example, the grid size may include the length of the unit grid in the horizontal direction and the vertical direction, and the like, and the number of grids may include the number of grids in the horizontal direction and the number of grids in the vertical direction, and the like. Of course, the attribute parameters of the model are not limited to the above examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the embodiments of the present disclosure, and all that can be achieved is intended to be covered by the scope of the embodiments of the present disclosure as long as the functions and effects achieved by the embodiments of the present disclosure are the same as or similar to the embodiments of the present disclosure.
In this embodiment, the number of horizontal grids, the number of vertical grids, and the specific values of the lengths of the unit grids in the horizontal and vertical directions may be set empirically, but the length of the entire grid in the horizontal direction is longer than the length of the horizontal well, the length of the entire grid in the vertical direction is longer than the length of the fracture perpendicular to the horizontal well, and the length of the shale gas horizontal well may be set reasonably. The limit adsorption concentration, langmuir pressure and critical desorption pressure may be set within a reasonable range with reference to the results of the article and experiment. It is sufficient that the pressure of the model constant pressure production is set to be less than the critical desorption pressure so that the gas can be desorbed. The attribute parameters of the model are not changed at the subsequent stage, only the geological parameters (such as porosity, permeability and the like) and the engineering parameters (such as the number of fracturing fracture sections and the half length of fractures) of the shale gas horizontal well are needed to be changed, a group of combinations (values of characteristic parameters) of the geological parameters and the engineering parameters are given, and the yield corresponding to the group of parameters can be obtained by utilizing the numerical simulation model. And then, history fitting is carried out by utilizing actual yield data, and the parameters are updated, so that the numerical simulation result can be used as the input of the long-short term memory neural network, and the long-short term memory neural network is utilized to predict the future moment, thereby ensuring the accuracy of the yield prediction model obtained by training under the condition of large time span.
In one embodiment, since natural gas in shale exists in three forms: free gas in rock pores, free gas in natural fractures, and adsorbed gas on the surface of organic minerals, these different reservoir mechanisms directly affect the manner, speed, and efficiency of shale gas development. The reservoir mechanism, the seepage mechanism and the development mode of the shale gas are the basis of the analysis of the productivity influence factors, so that a plurality of characteristic parameters influencing the productivity of the target shale gas horizontal well can be analyzed and determined from two aspects of geology and engineering.
In this embodiment, the geological parameter may be geological exploration data, which may be obtained by exploring or detecting a geological through various means and methods. In some embodiments, the geological parameters may include, but are not limited to, at least one of: shale reservoir thickness, formation pressure, reservoir properties.
In this embodiment, the engineering parameter may be engineering design data, and may be a development plan or design data that is made for the shale gas horizontal well before the development of the shale gas horizontal well. In some embodiments, the engineering parameters may include, but are not limited to, at least one of: horizontal well length, fracturing stage number, fracture half-length, fracture height, conductivity, and size of fracture modification Volume (SRV) area. Of course, it can also be understood that the engineering parameters may further include: the volume zone permeability is reformed through fracturing, the volume zone porosity is reformed through fracturing, the crack permeability, the number of cracks and the like are determined according to actual conditions, and the method is not limited.
In one embodiment, the characteristic parameters may be parameters that influence the productivity of the target shale gas horizontal well, which are analyzed and determined in advance from geological parameters and/or engineering parameters. In some embodiments, the plurality of characteristic parameters may include: reservoir thickness, initial formation pressure, matrix permeability, matrix porosity, fracture modification volumetric region permeability, fracture modification volumetric region porosity, fracture permeability, fracture half-length, and fracture number. Of course, the characteristic parameters are not limited to the above examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the embodiments of the present disclosure, and all such modifications are intended to be included within the scope of the embodiments of the present disclosure as long as they achieve the same or similar functions and effects as the embodiments of the present disclosure.
In this embodiment, the reservoir thickness, the initial formation pressure, the matrix permeability and the matrix porosity may belong to geological parameters, wherein the matrix permeability and the matrix porosity belong to reservoir properties. The permeability of the fracturing modification volume region, the porosity of the fracturing modification volume region, the permeability of cracks, the half length of the cracks and the number of the cracks can belong to engineering parameters.
In one embodiment, before obtaining the values of the plurality of characteristic parameters, the method may further include: the method comprises the steps of obtaining a plurality of geological parameters and a plurality of engineering parameters, and screening a plurality of parameters influencing the shale gas horizontal well productivity from the plurality of geological parameters and the plurality of engineering parameters by using a random forest characteristic importance analysis method. A plurality of parameters influencing the productivity of the shale gas horizontal well can be used as characteristic parameters, and the value range of each characteristic parameter is determined. Furthermore, a group of characteristic data sets can be generated according to the value range of each characteristic parameter; wherein the feature data set contains values of the respective feature parameters.
In the embodiment, because not every geological parameter and engineering parameter can affect the yield, a plurality of parameters affecting the shale gas horizontal well productivity can be screened from a plurality of geological parameters and a plurality of engineering parameters by using a random forest characteristic importance analysis method. The idea of evaluating the feature importance by using the random forest is as follows: and judging how much each feature contributes to each tree in the random forest, then taking an average value, and finally comparing the contribution of the features with the contribution of the features. The contribution may be calculated by an error rate of the out-of-bag data (data that is not decimated at the time of random sampling). First, for each decision tree in a random forest, calculating an out-of-bag data error using the corresponding out-of-bag data; secondly, noise interference is added to the characteristic X of all samples of the data outside the bag randomly, the value of the sample at the characteristic X is changed randomly, and the error of the data outside the bag is calculated again; the importance of the final random forest features is the average of all decision tree out-of-bag error differences. If the accuracy rate outside the bag is greatly reduced after noise is randomly added to a certain characteristic, the importance degree of the characteristic is higher. Based on the principle, all geological parameters and engineering parameters can be sequenced by using a random forest characteristic importance analysis method, and characteristic parameters influencing the shale gas horizontal well productivity are selected.
In the embodiment, the characteristic parameter values of different shale gas horizontal wells may be different, but each characteristic parameter of the shale gas horizontal well has a reasonable value range limited by actual geological conditions, so that the value range of each characteristic parameter can be determined. For example: the fracturing fluid comprises a reservoir thickness (15-100 m), an initial formation pressure (350-550 bar), a matrix permeability (0.000001-0.001 mD), a matrix porosity (0.054-0.08), a fracturing modification volume region permeability (0.1-10 mD), a fracturing modification volume region porosity (0.08-0.15), a fracture permeability (1000-50000 mD), a fracture half-length (60-280 m), a fracture number (10-30) and the like, wherein the bar (bar) is a pressure intensity unit, and the mD is millidarcy and a permeability unit. Of course, the value range of each characteristic parameter is not limited to the above examples, and other modifications may be made by those skilled in the art in light of the technical spirit of the embodiments of the present disclosure, but the functions and effects achieved by the embodiments of the present disclosure are all within the scope of the embodiments of the present disclosure.
In this embodiment, a group of feature data sets may be randomly generated according to the value ranges of the feature parameters, and in some embodiments, random Sampling may be performed according to the value ranges of the feature parameters by using a Latin Hypercube Sampling method (LHS), statistical Sampling, and the like, to generate model parameters of an initial set. Of course, the random sampling method is not limited to the above examples, and other modifications may be made by those skilled in the art in light of the technical spirit of the embodiments of the present disclosure, but the functions and effects achieved by the random sampling method are all covered by the scope of the embodiments of the present disclosure.
In one embodiment, generating a training data set based on the historical production data set for the target shale gas level well and the numerical simulation production data for the target shale gas level well over a preset time period may include: and generating a target yield data set according to the historical yield data set and the numerical simulation yield data in a preset time period. Acquiring a preset time step, and converting the yield data in the target yield data set into input and output formats of a long-short term memory neural network according to the preset time step to obtain a sample data set (because the long-short term memory neural network belongs to supervised learning, input and output are required to be given); the sample data set may include a plurality of groups of data, and each group of data includes yield data of a preset time step, numerical simulation yield data of a last time step in the preset time step, and yield data of a next time step. Further, a training data set, a verification data set and a test data set can be randomly generated according to a preset proportion according to the sample data set.
In this embodiment, when predicting a future time, the predicted output of the long-short term memory neural network may be used as the input of the next time step, and the numerical simulation yield data of the last time step in the preset time steps may be combined with the input data, where the numerical simulation yield data of the last time step in the preset time steps may be used as an independent input of the long-short term memory neural network, so as to effectively ensure the accuracy of the data for training.
In this embodiment, the sample data set may include multiple sets of sample data, and since the output at multiple times is used as input data and the output at the next time is used as output data, each set of sample data may include output data of a preset time step and output data of the next time step, where the output data at the next time step may be used as tag data to test the accuracy of the model training result.
In this embodiment, before performing model training, a plurality of groups of sample data in the sample data set may be randomly split into a training data set, a verification data set, and a test data set according to a preset ratio, where the verification set may not be set. The preset proportion may be a training data set: verifying the data set: the test data set = 3.
In one scenario example, a shale gas reservoir numerical simulation model may be built according to the following attribute parameters: the number of grids in the X direction was 150, the number of grids in the Y direction was 60, and the length of each unit grid in the X and Y directions was 10m. A shale gas horizontal well is arranged in the model, and the length of the shale gas horizontal well in the X horizontal direction is 1000m. The model simulates a shale gas reservoir by considering langmuir adsorption resolution with the limit adsorption concentration (mass of gas adsorbed per mass of solid) set at 0.0019362, the langmuir pressure set at 18.549bar and the critical desorption pressure set at 400bar. The model sets the constant pressure of 300bar for production, 15 years of shale gas reservoir production process is simulated, and the established shale gas reservoir numerical simulation model can be shown in figure 2.
In the present scenario example, a Latin Hypercube (LHS) method may be used to generate model parameters (including but not limited to reservoir thickness, initial formation pressure, matrix porosity, matrix permeability, etc.) of the initial set, and the historical production data of the target shale gas horizontal well is used to perform iterative fitting continuously until convergence, so as to update the model parameters in the initial set, and obtain a numerical simulation model after historical fitting and the production data of the target shale gas horizontal well in a preset time period.
In the present scenario example, at time t, or at time t step, the detailed calculation process for the long-short term memory neural network is as follows:
(a) At forgetting door f t Middle determination early unit state C t-1 Which information should be discarded in combination with the input information x t And the state h of the previous hidden layer t-1 By activating the function, a value between 0 and 1 is output to the cell state C t-1 0 means complete discard, 1 means complete retention;
f t ＝σ(W f x t +U f h t-1 +b f ) (1)
(b) Determining input x in input gate t Which information should be stored to the cell state C t And updating the input information i therein t And candidate cell state
i t ＝σ(W i x t +U i h t-1 +b i ) (2)
(c) Candidate unit state obtained according to input at time t
And C containing all the previous input information t-1 Updating the cell state C to obtain t time step t ；
(d) Using output information o in output gates t And cell state C t Confirmation output result h t ；
o t ＝σ(W o x t +U o h t-1 +b o ) (5)
h t ＝o t ·tanh(C t ) (6)
Wherein, W is an input weight, U is a recursion weight, b is a bias, and subscripts f, i and o respectively represent a forgetting gate, an input gate and an output gate; the active function sigma of a forgetting gate, an input gate and an output gate is a sigmoid function with a variable between 0 and 1; the activation function tanh is a hyperbolic tangent function that compresses the value between-1 and 1. The two activation functions are used to enhance the nonlinearity of the network, which can be expressed as:
in the scene example, the gradient disappearance problem of the RNN is solved by finely designing the long-term and short-term memory neural network structure and adjusting the state of the storage unit by the forgetting gate, the input gate and the output gate. Thus, for long term dependency problems in sequence data, long-short term memory neural networks can better capture and extract historical information and predict future development.
In this scenario example, for actual production data, assuming that there are T production data at time steps, arbitrarily taking a number T between 1 and T as a time step (T cannot be too large, it is reasonable to do), Q (1) to Q (T), and production data of S (T) may be used as first input data of the model (where Q (1) represents historical production data corresponding to the first time step, Q (T) represents historical production data corresponding to the T-th time step, S (T) represents numerically-simulated production data corresponding to the T time step, and Q (T + 1) is used as corresponding first output data; and taking the historical yield data of Q (2) to Q (T + 1) and the numerical simulation yield data of S (T + 1) as second input data of the model, and taking Q (T + 2) as corresponding second output data, wherein the input data and the output data exist, so that the long-short term memory neural network can be supervised and learned. And sequentially proceeding, so as to obtain a sample data set required by the model. Wherein, the proportion of the training set, the verification set and the test set can be 3. In some embodiments, the validation set may not be partitioned, and only the training set and the test set may be partitioned.
In the scenario example, when the yield data at the time of T +1 is predicted, the actual historical yield data of Q (T-T + 1) to Q (T) and the numerical simulation yield data of S (T) are input, and the long-short term memory neural network can predict the yield data at the time of T +1, which is predicted by the long-short term memory neural network and can be called as Q' (T + 1); the yield data at the time of predicting T +2 is historical yield data of Q (T-T + 2) to Q (T + 1) and numerical simulation yield data of S (T + 1), which need to be input, because the actual yield data of Q (T + 1) is not available (the actual yield data only reaches Q (T)), the result Q '(T + 1) of the long-short term memory neural network prediction needs to be used as Q (T + 1) to supplement the input data, and the data is input into the long-short term memory neural network, so that the yield data Q' (T + 2) at the time of T +2 can be predicted. By repeating the above steps, the yield at the future time can be predicted. The modeling process of the long-term and short-term memory neural network is as follows:
(1) And (5) data prediction processing. The raw data collected, i.e. the yield data, cannot be used directly and needs to be processed. Firstly, errors or even erroneous data, which may be called outliers (outliers), are generated during data acquisition or recording by an instrument, and are removed to reduce noise of the data, and then the data needs to be normalized. The long-short term memory neural network is sensitive to the size scale of data, the normalization of the data can avoid the influence of larger or smaller input parameters on the model, and the normalization formula is as follows:
wherein x is new The normalized data is obtained; x is the number of old Is the actual value of the data; x is the number of min Is the minimum value of the data; x is the number of max Is the maximum value of the data. The productivity prediction model obtained based on the training of the long-term and short-term memory neural network can be expressed as follows:
Q(t)＝F{S(t-1),Q(t-1),Q(t-2),…,Q(t-T)} (10)
wherein Q (t) is predicted daily oil production data of the tth time step; f is a prediction function obtained by learning and fitting; q (T-1), Q (T-2), \ 8230, Q (T-T) is the production data for the previous T time steps, and S (T-1) represents the production data for the numerical simulation at time T-1. Given input and output, the LSTM can be supervised to learn the laws between input and output. T represents the time window length also called time lag, if the time window size is 3, the input daily output data set of the LSTM long-short term memory neural network model needs to be rearranged into the following form:
(2) And (5) training the model. And (3) dividing the data preprocessed in the previous step into two parts, wherein the first 80% of the data is used as a training set, and the output value corresponding to the last 20% of the data is used as the verification of the model. (authentication set may not be divided)
(3) And training by using the established LSTM capacity prediction model. When the future time is predicted, because the daily output data after t time step is unknown, the prediction result of the LSTM long-short term memory neural network needs to be added into the prediction array of the next time step when predicting one time step later. Therefore, when the trained model is used for prediction, the input and output forms are as follows:
wherein, Q' (t + n) is the predicted value of the short-term memory neural network at the time step of t + n, and the output predicted values are all normalized values. After the output data is subjected to the denormalization, the daily oil production predicted value of the LSTM long-short term memory neural network at n time steps from t +1 to t + n can be obtained.
(4) And (6) evaluating the model. After the model is built, the accuracy and performance of the model need to be evaluated, and the model can be evaluated by using Root Mean Square Error (RMSE).
Wherein N is the total number of samples;
is the predicted value at the ith time, y i Representing actual production data at the ith time.
In this scenario example, if the RMSE does not meet the preset requirements, parameters in the long-short term memory neural network need to be adjusted to optimize the long-short term memory neural network. The parameters in the long-short term memory neural network may include: network layer number, LSTM unit number, regularization ratio, etc. The parameter tuning learning curves of the parameters in each long-short term memory neural network may be, as shown in fig. 3, the parameter tuning learning curves of the number of network layers, the number of LSTM units, and the regularization ratio (Dropout ratio).
In one embodiment, a yield decreasing curve drawn by using the training data set, the testing data set and the verification data set can be as shown in fig. 4, the abscissa is the number of data, and the ordinate is the gas production rate, wherein the original data is the yield of the next time step in each group of sample data, that is, the label data. Since there are five thousand groups of sample data in the sample data set in this embodiment, if the original data are all drawn, they will overlap with the prediction result of the yield prediction model, and for easier observation, the original data are only represented by limited points. According to fig. 4, it can be known that the yield predicted by the yield prediction model obtained by training the long-term and short-term memory neural network model is similar to the original data, that is, the prediction effect of the yield prediction model is good.
Based on the same inventive concept, the embodiment of the present specification further provides a shale gas yield determination apparatus based on a long-short term memory neural network, as in the following embodiments. The shale gas yield determination method based on the long and short term memory neural network is implemented by referring to the implementation of the shale gas yield determination method based on the long and short term memory neural network, and repeated parts are not described again. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated. Fig. 5 is a block diagram of a structure of a shale gas production determination apparatus based on a long-short term memory neural network according to an embodiment of the present disclosure, as shown in fig. 5, the shale gas production determination apparatus may include: the structure of the acquisition module 501, the preprocessing module 502, and the determination module 503 will be described below.
The obtaining module 501 may be configured to obtain yield data of the target shale gas horizontal well at each time in a period before the target time;
a preprocessing module 502, which may be configured to preprocess the yield data at each time in a period before the target time to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time;
a determining module 503, configured to determine, according to the target yield data set, a yield of the target shale gas horizontal well at the target time by using a yield prediction model obtained through pre-training; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment.
The embodiment of the present specification further provides an electronic device, and specifically, referring to the schematic structural diagram of the electronic device shown in fig. 6 based on the method for determining shale gas yield based on the long-term and short-term memory neural network provided in the embodiment of the present specification, the electronic device may specifically include an input device 61, a processor 62, and a memory 63. The input device 61 may be specifically configured to input production data of the target shale gas horizontal well at each time in a period before the target time. The processor 62 may be specifically configured to pre-process the yield data at each time in a cycle prior to the target time to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time; according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment. The memory 63 may be specifically configured to store parameters such as production of the target shale gas horizontal well at the target time.
In this embodiment, the input device may be one of the main devices for exchanging information between a user and a computer system. The input device may include a keyboard, a mouse, a camera, a scanner, a light pen, a handwriting input board, a voice input device, etc.; the input device is used to input raw data and a program for processing these numbers into the computer. The input device can also acquire and receive data transmitted by other modules, units and devices. The processor may be implemented in any suitable way. For example, the processor may take the form of, for example, a microprocessor or processor and a computer-readable medium that stores computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, an embedded microcontroller, and so forth. The memory may in particular be a memory device used in modern information technology for storing information. The memory may include multiple levels, and in a digital system, the memory may be any memory as long as it can store binary data; in an integrated circuit, a circuit without a physical form and with a storage function is also called a memory, such as a RAM, a FIFO and the like; in the system, the storage device in physical form is also called a memory, such as a memory bank, a TF card and the like.
In this embodiment, the functions and effects specifically realized by the electronic device may be explained by comparing with other embodiments, and are not described herein again.
There is further provided in an embodiment of the present specification a computer storage medium for a shale gas production determination method based on a long-short term memory neural network, the computer storage medium storing computer program instructions, which when executed, may implement: obtaining the yield data of the target shale gas horizontal well at each moment in the period before the target moment; preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time; according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment.
In the present embodiment, the storage medium includes, but is not limited to, a Random Access Memory (RAM), a Read-Only Memory (ROM), a Cache (Cache), a Hard Disk Drive (HDD), or a Memory Card (Memory Card). The memory may be used to store computer program instructions. The network communication unit may be an interface for performing network connection communication, which is set in accordance with a standard prescribed by a communication protocol.
In this embodiment, the functions and effects specifically realized by the program instructions stored in the computer storage medium can be explained by comparing with other embodiments, and are not described herein again.
It should be apparent to those skilled in the art that the modules or steps of the embodiments of the present specification described above can be implemented by a general purpose computing device, they can be centralized in a single computing device or distributed over a network of multiple computing devices, and alternatively, they can be implemented by program code executable by a computing device, so that they can be stored in a storage device and executed by a computing device, and in some cases, the steps shown or described can be executed in a different order therefrom, or they can be separately fabricated as individual integrated circuit modules, or multiple modules or steps therein can be fabricated as a single integrated circuit module. Thus, embodiments of the present description are not limited to any specific combination of hardware and software.
Although the embodiments herein provide the method steps as described in the above embodiments or flowcharts, more or fewer steps may be included in the method based on conventional or non-inventive efforts. In the case of steps where no necessary causal relationship logically exists, the order of execution of the steps is not limited to that provided by the embodiments herein. When the method is executed in an actual device or end product, the method can be executed sequentially or in parallel according to the embodiment or the method shown in the figure (for example, in the environment of a parallel processor or a multi-thread processing).
It is to be understood that the above description is intended to be illustrative, and not restrictive. Many embodiments and many applications other than the examples provided would be apparent to those of skill in the art upon reading the above description. The scope of the embodiments of the present description should, therefore, be determined not with reference to the above description, but should instead be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.
The above description is only a preferred embodiment of the embodiments of the present disclosure, and is not intended to limit the embodiments of the present disclosure, and it will be apparent to those skilled in the art that various modifications and variations can be made in the embodiments of the present disclosure. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the embodiments of the present disclosure should be included in the protection scope of the embodiments of the present disclosure.
Claims (7)
1. A shale gas yield determination method based on a long-term and short-term memory neural network is characterized by comprising the following steps:
obtaining the yield data of the target shale gas horizontal well at each moment in the period before the target moment; the previous period of the target time is a plurality of times of preset time step length before the target time;
preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set includes yield data at a plurality of time instants arranged in time;
according to the target yield data set, determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training; the yield prediction model is obtained based on long-term and short-term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment;
before determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training according to the target yield data set, the method further comprises the following steps:
setting attribute parameters of the model;
establishing a shale gas reservoir numerical simulation model according to the attribute parameters; the shale gas reservoir numerical simulation model is internally provided with a shale gas horizontal well;
acquiring a historical yield data set of the target shale gas horizontal well;
carrying out historical fitting on the shale gas reservoir numerical simulation model by utilizing the historical yield data set of the target shale gas horizontal well to obtain numerical simulation yield data of the target shale gas horizontal well in a preset time period;
generating a training data set according to the historical yield data set of the target shale gas horizontal well and the numerical simulation yield data of the target shale gas horizontal well in a preset time period;
and training the long-term and short-term memory neural network according to the training data set to obtain a target yield prediction model.
2. The method of claim 1, wherein generating a training data set from the historical production data set for the target shale gas level well and the numerically simulated production data for the target shale gas level well over a preset time period comprises:
generating a target yield data set according to the historical yield data set and numerical simulation yield data in a preset time period;
acquiring a preset time step;
converting the yield data in the target yield data set into an input/output format of the long-short term memory neural network according to the preset time step to obtain a sample data set; the sample data set comprises a plurality of groups of data, and each group of data comprises yield data of a preset time step, numerical simulation yield data of the last time step in the preset time step and yield data of the next time step;
and randomly generating a training data set, a verification data set and a test data set according to the sample data set according to a preset proportion.
3. The method of claim 1, wherein the attribute parameters of the model comprise: grid size, grid number, length of shale gas horizontal well in horizontal direction, langmuir pressure, limiting adsorption concentration, critical desorption pressure, pressure generated by constant pressure.
4. The method of claim 1, wherein pre-processing the production data at each time in a cycle prior to the target time to obtain a target production data set comprises:
converting the yield data of each moment in a period before the target moment into data which accords with the input format of the long-short term memory neural network according to a preset time step and is a target yield data set; and the time data contained in the previous period of the target time is equal to the preset time step.
5. A shale gas production determination apparatus based on a long-short term memory neural network, comprising:
the acquisition module is used for acquiring the yield data of the target shale gas horizontal well at each moment in a period before the target moment; the previous period of the target time is a plurality of times of preset time step before the target time;
the preprocessing module is used for preprocessing the yield data of each moment in a period before the target moment to obtain a target yield data set; wherein the target yield data set comprises yield data at a plurality of moments arranged in time;
the determining module is used for determining the yield of the target shale gas horizontal well at the target moment by utilizing a yield prediction model obtained by pre-training according to the target yield data set; the yield prediction model is obtained based on long-short term memory neural network training and used for predicting the yield of the target moment according to the yield data of a plurality of moments before the target moment;
wherein the determining module is further configured to:
setting attribute parameters of a model before determining the yield of the target shale gas horizontal well at the target moment by using a yield prediction model obtained by pre-training according to the target yield data set;
establishing a shale gas reservoir numerical simulation model according to the attribute parameters; the shale gas reservoir numerical simulation model is internally provided with a shale gas horizontal well;
acquiring a historical yield data set of the target shale gas horizontal well;
carrying out historical fitting on the shale gas reservoir numerical simulation model by utilizing the historical yield data set of the target shale gas horizontal well to obtain numerical simulation yield data of the target shale gas horizontal well in a preset time period;
generating a training data set according to the historical production data set of the target shale gas horizontal well and the numerical simulation production data of the target shale gas horizontal well in a preset time period;
and training the long-term and short-term memory neural network according to the training data set to obtain a target yield prediction model.
6. A shale gas production determination apparatus based on a long-short term memory neural network, comprising a processor and a memory for storing processor executable instructions which when executed by the processor implement the steps of the method of any of claims 1 to 4.
7. A computer readable storage medium having stored thereon computer instructions which, when executed, implement the steps of the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110096222.8A CN112761628B (en) | 2021-01-25 | 2021-01-25 | Shale gas yield determination method and device based on long-term and short-term memory neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110096222.8A CN112761628B (en) | 2021-01-25 | 2021-01-25 | Shale gas yield determination method and device based on long-term and short-term memory neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112761628A CN112761628A (en) | 2021-05-07 |
| CN112761628B true CN112761628B (en) | 2023-03-28 |
Family
ID=75707040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110096222.8A Active CN112761628B (en) | 2021-01-25 | 2021-01-25 | Shale gas yield determination method and device based on long-term and short-term memory neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112761628B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113338916A (en) * | 2021-07-16 | 2021-09-03 | 西南石油大学 | Method for predicting and diagnosing shaft effusion |
| CN113503158A (en) * | 2021-08-23 | 2021-10-15 | 重庆科技学院 | Prediction analysis method for relation between shale gas reservoir soaking time and yield |
| CN113962148B (en) * | 2021-10-20 | 2022-09-13 | 中国石油大学(北京) | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network |
| CN115860280B (en) * | 2023-02-27 | 2023-05-23 | 中国石油大学(华东) | Shale gas yield prediction method, device, equipment and storage medium |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106351651B (en) * | 2016-08-26 | 2019-10-11 | 中国石油天然气股份有限公司 | The prediction technique and device of shale gas well deliverability |
| CN108661631B (en) * | 2017-04-01 | 2021-06-15 | 中国石油化工股份有限公司 | Yield prediction method |
| CN108664678B (en) * | 2017-04-01 | 2021-08-31 | 中国石油化工股份有限公司 | Yield prediction method |
| CN109214026A (en) * | 2017-07-07 | 2019-01-15 | 中国石油天然气股份有限公司 | A kind of shale gas horizontal well initial productivity prediction technique |
| CN108240214A (en) * | 2018-01-18 | 2018-07-03 | 西南石油大学 | PRODUCTION FORECASTING METHODS after a kind of shale gas reservoir horizontal well fracturing pressure |
| CN108595803B (en) * | 2018-04-13 | 2021-04-23 | 中国石油化工股份有限公司 | Shale gas well production pressure prediction method based on recurrent neural network |
| CN109472088B (en) * | 2018-11-12 | 2021-02-23 | 中国石油化工股份有限公司 | Shale gas-conditioned production well production pressure dynamic prediction method |
| CN110533163A (en) * | 2019-08-02 | 2019-12-03 | 中国石油集团长城钻探工程有限公司 | Sand plugging after fracturing method for early warning based on long Memory Neural Networks in short-term |
| CN112228054B (en) * | 2020-11-06 | 2022-11-08 | 中国石油大学(北京) | Method, device and equipment for determining shale gas yield based on convolutional neural network |
-
2021
- 2021-01-25 CN CN202110096222.8A patent/CN112761628B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112761628A (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112761628B (en) | Shale gas yield determination method and device based on long-term and short-term memory neural network | |
| CN112228054B (en) | Method, device and equipment for determining shale gas yield based on convolutional neural network | |
| CN106407649B (en) | Microseismic signals based on time recurrent neural network then automatic pick method | |
| CN109034194B (en) | Transaction fraud behavior deep detection method based on feature differentiation | |
| CN112989708B (en) | Well logging lithology identification method and system based on LSTM neural network | |
| CN110442666B (en) | Mineral resource prediction method and system based on neural network model | |
| CN111832227B (en) | Shale gas saturation determination method, device and equipment based on deep learning | |
| CN103617147A (en) | Method for identifying mine water-inrush source | |
| CN110851176B (en) | Clone code detection method capable of automatically constructing and utilizing pseudo-clone corpus | |
| CN112016097B (en) | Method for predicting network security vulnerability time to be utilized | |
| CN111048163B (en) | Shale oil hydrocarbon retention amount (S1) evaluation method based on high-order neural network | |
| CN112712025A (en) | Complex lithology identification method based on long-term and short-term memory neural network | |
| CN110807544A (en) | Oil field residual oil saturation distribution prediction method based on machine learning | |
| CN115659254A (en) | Power quality disturbance analysis method for power distribution network with bimodal feature fusion | |
| Chu et al. | Co-training based on semi-supervised ensemble classification approach for multi-label data stream | |
| CN112035345A (en) | Mixed depth defect prediction method based on code segment analysis | |
| CN115062727A (en) | Graph node classification method and system based on multi-order hypergraph convolutional network | |
| CN113962148B (en) | Yield prediction method, device and equipment based on convolutional coding dynamic sequence network | |
| CN110260914A (en) | A kind of project security monitoring system realm division methods based on measuring point space-time characteristic | |
| CN115620068A (en) | Rock lithology automatic identification and classification method under deep learning mode | |
| Chen | Hotel management evaluation index system based on data mining and deep neural network | |
| Cirac et al. | Deep hierarchical distillation proxy-oil modeling for heterogeneous carbonate reservoirs | |
| Li et al. | A novel hybrid CNN–SVM method for lithology identification in shale reservoirs based on logging measurements | |
| CN117235628B (en) | Well logging curve prediction method and system based on hybrid Bayesian deep network | |
| Yang et al. | Artificial Intelligence In Source Discrimination of Mine Water: A Deep Learning Algorithm For Water Source Discrimination |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |